Autopilot-Preserving Residual Q-Learning with HJB-Inspired Finite-Action Risk Filtering for Fixed-Wing UAV Command Supervision
arXiv:2606.01397v1 Announce Type: cross Abstract: A fixed-wing UAV must hold airspeed, altitude, and heading references under wind, gusts, and turbulence, channels coupled so that correcting one can degrade another. Classical autopilots stabilize the airframe well but adapt poorly when a hard crosswind meets an aggressive turn, while reinforcement-learning (RL) policies acting directly on the surfaces concentrate exploration risk at the actuator interface. We place a learned supervisor above an unchanged autopilot rather than inside it: it selects a residual from a finite, bounded action set o
The continuous development in reinforcement learning (RL) combined with demand for more robust autonomous systems drives innovation in UAV control. This research leverages existing autopilot stability with a learned supervisory layer, indicating a current trend towards modular AI integration.
This development offers a practical approach to enhancing UAV autonomy and reliability, critical for both commercial and defense applications by mitigating risks associated with full RL control while improving performance in complex conditions. This advancement contributes to safe and effective deployment of AI in physical systems.
UAV control systems can now integrate advanced AI supervisors without fully replacing existing, stable autopilots, potentially accelerating the development and adoption of more resilient autonomous flight. This reduces the barriers and risks associated with deploying new AI models directly into critical control loops.
- · Defence contractors
- · UAV manufacturers
- · Logistics companies (drone delivery)
- · AI/ML research institutions
- · Developers of fully end-to-end RL for physical control (if safety concerns persi
Increased operational capability and safety for fixed-wing UAVs in challenging environmental conditions.
Reduced testing and certification timelines for autonomous UAVs due to the lower risk profile of a supervisory AI approach.
Broader adoption of autonomous fixed-wing UAVs for various tasks, including reconnaissance, cargo, and infrastructure inspection, due to enhanced reliability.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG